Most tech giants hope to sell AI to the masses. But Mark Zuckerberg is giving away what Meta considers one of the world’s best AI models for free.
Meta released the largest, most proficient version of a enormous language model called Llama on Monday, free of charge. Meta did not disclose the cost of developing Llama 3.1, but Zuckerberg he recently told investors that his company spends billions on the development of artificial intelligence.
With this latest release, Meta is showing that the closed approach favored by most AI companies isn’t the only way to develop AI. But the company is also putting itself at the center of a debate about the dangers of releasing AI unchecked. Meta trains Llama in a way that prevents the model from generating harmful results by default, but the model can be modified to remove such safeguards.
Meta claims that Llama 3.1 is as intelligent and useful as the best commercial offerings from companies like OpenAI, Google, and Anthropic. In some benchmarks measuring progress in AI, Meta claims that the model is the smartest AI on Earth.
“It’s very exciting,” he says. Percy Liangassociate professor at Stanford University who tracks open-source AI. If developers find the novel model as capable as industry leaders, including OpenAI’s GPT-4o, Liang says many could switch to Meta’s offering. “It’ll be interesting to see how adoption changes,” he says.
IN open letter published with the release of the novel model, CEO Meta Zuckerberg compared Llama to the open-source Linux operating system. When Linux took off in the tardy 1990s and early 2000s, many major technology companies invested in closed-source alternatives and criticized open-source software as risky and unreliable. But Linux is now widely used in cloud computing and is the core of the Android mobile operating system.
“I believe AI will evolve in a similar way,” Zuckerberg writes in his letter. “Currently, a handful of technology companies are developing leading closed models. However, open source software is rapidly closing the gap.”
But Meta’s decision to give away its AI isn’t without self-interest. Previous Llama releases have helped the company secure a position of influence among AI researchers, developers, and startups. Liang also notes that Llama 3.1 isn’t truly open source, because Meta places restrictions on its operate, such as limiting the scale at which the model can be used in commercial products.
The novel version of Llama has 405 billion parameters or modifiable elements. Meta has already released two smaller versions of Llama 3, one with 70 billion parameters and the other with 8 billion. Meta has also released improved versions of these models today under the Llama 3.1 brand.
Llama 3.1 is too enormous to run on a regular computer, but Meta says a number of cloud providers, including Databricks, Groq, AWS, and Google Cloud, will offer hosting options to allow developers to run custom versions of the model. The model can also be accessed at Meta.ai.